ARTICLE

Received 4 Aug 2015 | Accepted 18 Dec 2015 | Published 3 Feb 2016 DOI: 10.1038/ncomms10516 OPEN Marine mammals harbor unique microbiotas shaped by and yet distinct from the sea

Elisabeth M. Bik1,2, Elizabeth K. Costello1, Alexandra D. Switzer1, Benjamin J. Callahan3, Susan P. Holmes3, Randall S. Wells4, Kevin P. Carlin5, Eric D. Jensen6, Stephanie Venn-Watson5 & David A. Relman1,2,7

Marine mammals play crucial ecological roles in the oceans, but little is known about their microbiotas. Here we study the bacterial communities in 337 samples from 5 body sites in 48 healthy dolphins and 18 healthy sea lions, as well as those of adjacent seawater and other hosts. The bacterial taxonomic compositions are distinct from those of other mammals, dietary fish and seawater, are highly diverse and vary according to body site and host species. Dolphins harbour 30 bacterial phyla, with 25 of them in the mouth, several abundant but poorly characterized Tenericutes species in gastric fluid and a surprisingly paucity of Bac- teroidetes in distal gut. About 70% of near-full length bacterial 16S ribosomal RNA sequences from dolphins are unique. Host habitat, diet and phylogeny all contribute to variation in marine mammal distal gut microbiota composition. Our findings help elucidate the factors structuring marine mammal microbiotas and may enhance monitoring of marine mammal health.

1 Department of Microbiology and Immunology, Stanford University School of Medicine, Stanford, California 94305, USA. 2 Veterans Affairs Palo Alto Health Care System, Palo Alto, California 94304, USA. 3 Department of Statistics, Stanford University, Stanford, California 94305, USA. 4 Sarasota Dolphin Research Program, Chicago Zoological Society, c/o Mote Marine Laboratory, Sarasota, Florida 34236, USA. 5 Translational Medicine and Research Program, National Marine Mammal Foundation, San Diego, California 92106, USA. 6 Space and Naval Warfare Systems Center Pacific, San Diego, California 92152, USA. 7 Department of Medicine (Infectious Diseases and Geographic Medicine), Stanford University School of Medicine, Stanford, California 94305, USA. Correspondence and requests for materials should be addressed to D.A.R. (email: [email protected]).

NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516

arine mammals play essential roles in the marine Results ecosystem as apex predators, primary and secondary A survey of marine mammal microbiota taxonomic composition. Mconsumers, and indicators of ocean health1–3. Despite We examined bacterial phylogenetic diversity (PD) in the mouth, their importance, man has driven many marine mammal forestomach fluid (‘gastric fluid’), respiratory tract (blowhole species to endangered status, and even extinction by hunting, swab and forceful exhalation referred to as chuff) and rectum of overfishing and exploitation of their marine environment. 38 common bottlenose dolphins (Tursiops truncatus; order Although conservation efforts have allowed some species to Cetacea) and in the mouth, stomach and rectum of 18 California recover, many are still threatened, and there is an increasing need sea lions (Zalophus californianus; order Carnivora), from a group to better understand chronic, subclinical effects of ocean-borne of healthy animals under the care of the US Navy Marine threats on marine mammals and their populations’ long-term Mammal Program (MMP) in San Diego Bay, San Diego, CA survival and reproductive success4. (Supplementary Table 1), as well as the fish and squid that Mammal-associated microbial communities play critical roles comprised their diet at this facility. In addition, we examined in host nutrition, differentiation of host tissues, colonization bacterial diversity in oral and rectal specimens from 10 healthy, resistance and the development of immune function, as well as wild bottlenose dolphins during capture-release health other beneficial processes. Studies of faecal specimens from 60 assessments in Sarasota Bay, Florida3. Seawater specimens were terrestrial mammalian species have shown a striking degree of collected adjacent to each animal at the time of animal sampling, host specificity of microbiota, reflecting the influence of host and processed and analysed together with the animal specimens. phylogeny, gut anatomy, and diet5,6. Despite a growing Using 2 different complementary amplification and sequencing understanding of the factors that shape the structure and approaches, a total of 20,030 near-full length (FL) function of the microbiotas of terrestrial animals, relatively little Sanger-sequenced bacterial 16S ribosomal RNA (rRNA) gene is known about those that influence the microbiotas of marine amplicons and 915,150 pyrosequencing (PS) reads from the mammals. V3–V4–V5 region of the bacterial 16S rRNA were analysed, Marine mammals provide an unusual opportunity to explore providing both high taxonomic resolution (FL sequences) as well the relatedness of mammalian microbiotas as a function as deeper community coverage (PS reads). of evolutionary divergence. Extant marine mammal species descended from terrestrial ancestors that recolonized the sea and belong to one of three orders: Cetacea (whales, dolphins High diversity within marine mammal microbiotas. Of all and porpoises), Sirenia (manatees and dugongs) or Carnivora samples included in our study, seawater samples contained the (sea lions, seals, walruses, sea otters and polar bears). highest phylum-level bacterial diversity: 48 phyla were detected Cetaceans are derived from the group Cetartiodactyla7; the with PS (Fig. 1, Supplementary Figs 1 and 2, Supplementary hippopotamus family (Hippopotamidae) represents the cetaceans’ Data 1). Yet, representatives of 30 bacterial phyla were detected in closest extant relatives8–10. Unlike hippopotamuses, cetaceans the dolphin and sea lion microbiota specimens from the 5 body are carnivorous, and the diet of their most recent, now sites, based on PS reads (Fig. 1, Supplementary Fig. 1); 22 of these extinct, common ancestor is unknown9,11. The Carnivora were also detected in the FL data set (Supplementary Figs 3 and suborder, Pinnipedia (sea lions, seals and walruses) is most 4a). All 30 phyla were found in the dolphin specimens alone, with closely related to the musteloids (raccoons, red pandas, skunks oral, gastric fluid and chuff specimens containing the highest and weasels), with ursids (bears) as their sister group. Like number of phyla (Supplementary Data 1). Most animal-asso- their closest extant relatives, pinnipeds are carnivorous12. ciated and seawater specimens were dominated by Members of Sirenia are most closely related to the Proboscidea and Bacteroidetes, except for dolphin rectal specimens, which (elephants), and both groups share a herbivorous diet. Although contained surprisingly few Bacteroidetes sequences (see below) these marine mammal orders are phylogenetically distinct and (Supplementary Fig. 2). Dolphin gastric fluid specimens con- their members exploit different food sources, all live and feed in tained high relative proportions of Tenericutes, a phylum usually an aquatic environment and therefore share many convergent found at low abundance in mammal-associated microbiotas. adaptations10. Representatives from 13 candidate phyla (phyla for which there Several recent studies have explored marine mammal micro- are no laboratory-cultivated isolates) were detected in PS data biotas, and provided important early assessments of microbial from the marine mammal microbiotas (Fig. 1, Supplementary diversity in these critical host species13–16. Some culture-based Fig. 1). Of these, GN02 was the most abundant, and was found studies have characterized gastrointestinal microbiotas of primarily in dolphin oral (0.3% of PS reads), gastric (0.3%) and cetaceans17–19. Small scale, culture-independent studies have respiratory specimens (1.2%). About 1.9% of the dolphin oral characterized the respiratory microbiota of dolphins15,20. Apprill reads were assigned to phylum H-178; all of these reads were et al.21 surveyed the skin microbiota of baleen whales. Distal gut detected in the MMP dolphins. This phylum was not found in the microbiotas of seals16,22,23, dugong24, finless porpoise25 and oral specimens of wild dolphins from Sarasota Bay, or in any of baleen whales26 have also been surveyed. One study compared the the sea lion specimens. microbiotas of terrestrial mammals with those of two species of Seawater samples displayed the highest Operational Taxo- seals and a dugong14; their results suggested that marine nomic Unit (OTU; species level) richness. Of the marine mammals harbour gut microbiotas that are more species rich mammal-associated specimens, dolphin oral and respiratory than their terrestrial counterparts. specimens showed the highest number of observed OTUs, and In this study, we analyse bacterial diversity in 337 specimens dolphin rectal specimens the lowest (Supplementary Fig. 5). from 5 different body sites in 48 dolphins (order Cetacea) and 18 Among the marine mammal-associated specimens, the largest sea lions (order Carnivora), as well as adjacent seawater and gain of PD (that is, the branch length a sample adds to a tree dietary fish, and then compare our data with those from containing other samples) was found with the dolphin oral previously published studies of other animals. Our findings specimens, even when sea lion and other dolphin specimens were expand our understanding of bacterial diversity and microbial considered first (Fig. 2). Gains of PD were much lower in dolphin community structure in mammals and the influence of life in the rectal and respiratory specimens, and sea lion gastric fluid and sea, as well as enhance the means for monitoring and improving rectal specimens. Seawater samples yielded the highest gain in PD, the health of marine mammals. even after addition of all marine mammal-associated sequences.

2 NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516 ARTICLE

Dolphins Sea lions

OralGastric Rectal Chuff BlowOral Gast Rect Seawater All F&S 100% MMW M W MMMMMMW Other phyla Chlorobi KSB3 WS3 80% Synergistetes Chloroflexi SR1 Planctomycetes 60% SAR406 OD1 Verrucomicrobia H-178 40% GN02 Spirochaetes Relative abundance Actinobacteria Cyanobacteria 20% Tenericutes Firmicutes Fusobacteria Bacteroidetes Proteobacteria 1 1 3 5 7 8 9 1 2 6 1 4 1 1 1 15 12 13 10 16 16 17 18 19 22 26 32 38 50 60 66 12 15 10 72 76 78 18 20 21 27 28 30 31 33 35 36 37 39 40 41 42 43 45 46 47 48 49 51 52 53 55 56 57 58 59 61 62 65 13 82 88 98 68 70 71 77 17 19 29 63 80 81 83 85 86 87 89 90 91 92 93 95 96 97 99 23 25 67 69 73 75 79 14 14 34 44 54 64 24 74 84 94 1 1 2 1 1 1 1 1 1 1 1 167 171 173 135 138 139 148 163 177 183 185 189 208 210 212 216 128 131 165 166 168 169 170 172 136 137 140 141 142 143 145 146 147 149 150 151 152 153 155 156 157 158 159 160 161 162 218 220 222 226 175 176 178 179 180 181 182 186 187 188 190 191 192 193 195 196 197 198 199 200 201 202 203 205 206 207 209 213 215 217 120 121 122 123 125 126 127 129 130 132 133 219 221 223 225 227 101 102 103 105 107 108 100 106 109 144 154 174 204 214 124 134 224 184 164 104 0% 194

Figure 1 | Bacterial phyla in specimens from 38 dolphins and 18 sea lions. Only one time point per animal and only specimens with Z372 pyrosequencing reads are shown (n ¼ 199, average number of reads per specimen was 3119.4). Four other single time point specimens and all 26 extraction controls yielded o50 reads. Specimens are shown in the same order as listed in Supplementary Table 1. The relative proportions of each phylum within each specimen and in the total data set are shown sorted on the average abundance in the combined data set (most abundant at the bottom). A total of 51 phyla were found. Only the 20 most abundant phyla are shown; the remaining 31 as well as all unclassifiable sequences were grouped together in ‘other phyla’; these are shown in Supplementary Fig. 1. Of the 20 phyla shown here, 19 were found in specimens obtained from the marine mammals; phylum SAR406 was only found in seawater samples. The horizontal lines at the top show the specimen source. ‘M’ and ‘W’ indicate specimens obtained from MMP and wild dolphins, respectively. No gastric or respiratory specimens were obtained from the wild dolphins, and in several cases different MMP dolphins were used for the respiratory sampling than for the oral/gastric/rectal sampling. Column ‘All’ on the right displays the average relative phyla abundance in all specimens. Blow, blowhole; Gast, gastric; F&S, fish and squid.

ab 1 1 Dolphin oral Sea lion rectal 0.9 0.9 Dolphin gastric Sea lion gastric 0.8 0.8 Dolphin rectal Sea lion oral 0.7 0.7 Dolphin chuff Dolphin blowhole 0.6 0.6 Dolphin blowhole Dolphin chuff 0.5 0.5 Sea lion oral Dolphin rectal 0.4 0.4 Sea lion gastric Dolphin gastric 0.3 0.3 Sea lion rectal Dolphin oral 0.2 0.2 Fish and squid Fish and squid 0.1 0.1 Seawater Seawater Cumulative phylogenetic diversity Cumulative phylogenetic diversity 0 0 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Number of reads (× 100,000) Number of reads (× 100,000)

Figure 2 | Gain in phylogenetic diversity according to specimen type. Phylogenetic diversity (PD) gain refers to the cumulative branch length in a phylogenetic tree that is exclusive to a particular specimen, as samples are added in a user-defined order. Calculations were done on the PS data set. Two different orders are shown. In a, the dolphin oral specimens were added first, followed by dolphin gastric, rectal and respiratory samples, sea lion specimens, fish and squid, and seawater samples. Panel b shows the animal samples in reversed order. Only one time point per animal was included. Of the marine mammal specimens, the dolphin oral specimens show the most phylogenetic gain, even when the sea lion specimens are considered first.

The FL sequence data set allowed a high taxonomic novel (Supplementary Data 2). These percentages are noteworthy resolution analysis of bacterial novelty found in these marine in comparison to early studies of the human microbiota27–29 mammals. The dolphin-associated FL sequences had a (Supplementary Fig. 6). surprisingly high degree of novelty; overall, 70% of the dolphin-associated FL-OTUs (36% of sequences) and 44% of sea lion FL-OTUs (31% of sequences) were not similar to Microbiotas from marine mammals and seawater are distinct. previously published sequences at a 97% sequence identity One possible explanation for the high degree of bacterial novelty cutoff, with the greatest amount of novelty found in the in the two marine mammal species is their sharing or acquisition dolphin oral cavity (Supplementary Fig. 6). At 90% identity, of bacterial taxa from seawater. To address this possibility, we 4.0% of marine mammal-associated OTUs could be considered compared the bacterial communities of sea lions and dolphins

NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516

abDolphins Sea lions

OralGastric Rectal Chuff BlowOral Gast RectF&S Seawater MMW M W MMMMMMW Bacteroides OTU 556491 1,30470 551 Cetobacterium 12 OTU 11302 66 26

Pelagibacter 2,108 OTU 345876

Abundance

0.25000

0.01563

0.00098

Cardiobacteriales c OTU 557712 9 Pelagibacter ubique 9 Fusobacterium sp. C104 8 Tenericutes sp.A476 9 12 7 sp. B031 20 5 6 13 Tenericutes 5 4 OTU 2329431 4 or swab 9 4 1 3

Fusobacterium 2 13 1 OTU 809380 1 2 0 0 0 Log 16S rRNA gene copies per ml 0 Seawater Oral Gastric Rectal OTU 1124928 N =13 N =20 N =9 N =13 Figure 3 | Habitat-specificity of bacterial groups found in this study. (a) Relative OTU abundance in the pyrosequencing data set. Specimens with Z372 reads (n ¼ 199; single time point per animal; same specimens and order as in Fig. 1) are shown in columns, and OTUs with Z20 reads (n ¼ 236) are shown in rows. The OTUs are clustered using an NMDS sorting based on Bray–Curtis distance, while the specimens (shown in columns) are sorted per specimen group. Relative abundance is shown in grey scale (white, absent; black, high abundance). (b) Venn diagram showing sharing of the 4,137 OTUs found in a rarefied pyrosequencing data set of 18 dolphins, 18 sea lions and 18 water specimens. Each data set was rarefied to 69,377 reads to match the smallest data set (from the water samples). Green, dolphins (1,452 OTUs total, includes oral, gastric, and rectal specimens from MMP and wild dolphins); red, sea lions (659 OTUs oral, gastric, rectal); blue, seawater (2,212 OTUs). The overlap between the circles is not to scale. (c) Quantitative PCR on four specimen types. Results are presented as average gene copy number per millilitre for seawater and gastric fluid, or per swab for oral and rectal specimens, for each of four different qPCR tests. Each individual qPCR was done in triplicate, error bars indicate s.d.. The numbers above the bars indicate the number of specimens positive for each qPCR test. with the bacterial composition of seawater adjacent to these found in seawater (Fig. 3b), based on a rarefied data set with animals at the time of their sampling. equal numbers of PS reads and specimens representing In both sequencing approaches, the bacterial community dolphins, sea lions and water. The distinctness of the bacterial composition in seawater was profoundly different from that in communities found in seawater and in dolphins was the marine mammals (Fig. 3a, Supplementary Fig. 2). Despite the confirmed by quantitative PCR (qPCR) assays developed to dominance of the phylum Proteobacteria at most body sites in detect bacterial groups highly abundant in each of three dolphin dolphins and sea lions, as well as in sea water, Gammaproteo- body sites and seawater (Fig. 3c). For example, Pelagibacter were much more common in dolphins and sea lions, and ubique, an abundant bacterium in most ocean surveys, was much more common in seawater present at 10-fold lower concentrations in dolphin gastric fluid (Supplementary Fig. 4b). In particular, the 10 most abundant than in seawater collected next to the animals, and at 4100-fold seawater-associated OTUs found by PS (PS-OTUs) together lower concentrations in dolphin oral specimens. Dolphin accounted for 39.1% of all seawater reads but for only 0.06% of stomach and rectum-associated Actinobacillus sequences were the marine mammal-associated sequences (Supplementary detected at 1–10 million-fold lower abundances in seawater Data 3). Most (278 out of 437) of the seawater-associated reads (Fig. 3c). in marine mammal specimens were found in dolphin To investigate other potential sources of microbial diversity in forestomachs, suggesting that some dolphins might have ingested marine mammals, we analysed the microbiotas of the fish and seawater just before or at the time of sample collection. While the squid that are fed to the MMP dolphins and sea lions, but found 10 most abundant dolphin rectum-associated PS-OTUs these specimens contained distinct bacterial communities as well accounted for 81.1% of the dolphin rectum reads, their relative (Figs 1 and 3a, Supplementary Figs 2 and 4a). None of the abundance in seawater collected next to these animals was only fish- and squid-associated bacterial sequences were abundant in 0.05% (Supplementary Data 3). the marine mammals. The 10 most abundant PS-OTUs in fish About 95.3% of the bacterial PS-OTUs in seawater were and squid specimens together comprised 72.3% of reads from not found in dolphins or sea lions, and 94.9% of the these specimens, but only 0.02% of all marine mammal-associated PS-OTUs found in the two marine mammals species were not reads (Supplementary Data 3).

4 NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516 ARTICLE

a Location b 2 San Diego Alternate locations Sarasota Bay Specimen type 1 Dolphin oral Dolphin gastric Dolphin rectal Distance Bik et al. Figure 5 Dolphin chuff 0.0 0 Dolphin blowhole 0.1 NMDS2 Sea lion oral 0.2 Sea lion gastric 0.3 Sea lion rectal 0.4 −1 Fish and squid 0.5 Seawater San Diego Seawater alternate Seawater Sarasota −2 −1 0 1 2 NMDS1

Figure 4 | Relationships among bacterial communities from different host species and habitats. Only samples with Z372 reads were included (n ¼ 199; one specimen per body site per animal). (a) NMDS Bray–Curtis ordination of PS-analysed specimens from dolphins and sea lions included in this study. Colours indicate the different specimen types and sources. Shapes display the location of the animal during sampling (San Diego or an undisclosed alternate, distant location for the MMP animals and Sarasota Bay for the wild, free-living dolphins). (b) Distance Threshold Network analysis of marine mammal-associated bacterial communities. The network displays binary relationships between specimens and OTUs using a Bray–Curtis distance and a Fruchterman–Reingold layout. Each data point represents the bacterial community from an individual specimen. Two specimens are considered ‘connected’ if the distance between them is less than a user-defined threshold, here, 0.6. The thickness of the connecting edge is related to the distance between two specimens. OTU abundances were proportionally transformed.

Microbiotas differ among host species and body sites. The 2329431 (40.2% of forestomach PS reads) was found in composition of the indigenous microbiotas of dolphins and sea almost all (17 of 18) dolphin gastric fluid specimens. It is lions varied as a function of body site and host species (Fig. 4a, identical to a sequence found in the stool of the Yangtze finless Supplementary Fig. 7). PS and FL read analysis portrayed very porpoise25. The closest relatives of this sequence type (89% similar results in terms of the most abundant bacterial groups sequence identity) were found in the genus Ureaplasma. qPCR found among the sampled marine mammal body sites analysis revealed this sequence type in the forestomach of all (Supplementary Fig. 2). Network analysis supported the role of dolphins in high abundance (107 copies per ml gastric fluid), but host species and body site in explaining variance in community not in dolphin oral specimens, sea lion-derived specimens or composition, and suggested close relationships between dolphin seawater (Fig. 3c). This sequence comprised 27.8% of dolphin chuff and blowhole specimens, and between sea lion oral and forestomach PS reads (in 56 out of 69 gastric samples), 0.51% of gastric specimens (Fig. 4b). Interestingly, seawater bacterial dolphin rectal reads, o0.01% of dolphin oral, sea lion oral and communities from different oceans were more similar to each gastric reads, and was not found among a total of 383,139 dolphin other than were communities from the same body site in different respiratory, sea lion rectal, fish, squid and seawater reads marine mammal species living in close proximity. No connections (Supplementary Data 3). between the seawater samples and any of the marine mammal The microbial communities found in dolphin rectal specimens body sites were found. contained half as many bacterial phyla and four times fewer No effect of age (2–51 years old) or sex on the composition of bacterial species as the dolphin oral specimens (Fig. 1, the dolphin microbiotas was apparent in ordinations, irrespective Supplementary Figs 2 and 5). Among the 12 bacterial phyla of sequencing approach (Supplementary Fig. 8). detected in the rectal specimens, Firmicutes, Proteobacteria and The dolphin mouth showed high bacterial richness Fusobacteria were most abundant based on both sequencing (Supplementary Fig. 5) and the highest PD (Fig. 2) and novelty approaches. The most abundant FL-OTU (29.3%) in dolphin (Supplementary Fig. 6). We detected 25 bacterial phyla in these rectal swabs, OTU_B031, was closely related to Actinobacillus specimens, with a predominance of Bacteroidetes and sequences previously detected in cetaceans25,31. qPCR assays Proteobacteria, but a remarkably high number of additional using specific primers detected this sequence in all gastric fluid phyla, including a surprising number of candidate phyla (n ¼ 11, specimens from both sea lions and dolphins (Fig. 3c) at high 2.7% of PS reads) (Fig. 1, Supplementary Figs 1 and 2, abundances, and in 4 of 20 oral specimens at approximately Supplementary Data 1). In the FL data set with higher taxonomic 10,000-fold lower abundances. A FL-OTU corresponding to resolution, OTU_C214 (Family Moraxellaceae, Class Gamma- Cetobacterium ceti32, OTU_K184, was the second most abundant proteobacteria) was the most abundant (7.1%) and prevalent (12.6%) among FL sequences and the most abundant (30.7%) (found in 21 of 22 dolphins). Another abundant FL-OTU (3.6%) among PS reads from dolphin rectal swabs (Supplementary was C104 (95.0% identical to Fusobacterium nucleatum). This Data 3). Of note, distinct Arthromitus sequences were found in 49 FL-OTU was found in 11 of 22 dolphin oral specimens and in all and 10% of dolphin and sea lion rectal specimens (via PS), of 20 dolphin oral samples by qPCR (Fig. 3c). In addition, our respectively. Bacteria assigned to this genus in the study confirms the presence of a specific group of oral Firmicutes phylum are also called segmented filamentous Campylobacter-like sequences in cetaceans30. bacteria (SFB). They have been commonly detected in the Twenty-two bacterial phyla were found in the dolphin gut communities of both invertebrates and vertebrates, with forestomach, of which Tenericutes, Bacteroides and different Arthromitus clades detected in different hosts33.An Proteobacteria were the most dominant (Supplementary Fig. 2). analysis of the FL Arthromitus sequences found in our study The most abundant FL-OTU, OTU_A476 in phylum Tenericutes confirmed the host specificity of these indigenous gut bacteria (36.4% of forestomach-derived FL sequences) or PS-OTU (Supplementary Fig. 9).

NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516

The bacterial communities of the two respiratory dolphin mammals13, for example, Brucella spp., Erysipelothrix spp. and specimen types, chuff and blowhole swab, were very similar to Streptococcus group D, were detected in dolphins or sea lions, but each other, and dominated by Proteobacteria and Bacteroidetes. Staphylococcus aureus was detected in multiple gastric fluid However, the chuff-derived PS reads were affiliated with a greater samples from dolphin Y and one from dolphin X. Several number of phyla (21) than a similar number of blowhole reads (11), sequence types within the genus Leptospira were detected, but as well as with more OTUs (Supplementary Figs 2 and 5). The none close to Leptospira interrogans, a known pathogen in dominant bacterial taxa in the bottlenose dolphin blowhole and pinnipeds. One sea lion (SL08) carried an oral Mycoplasma chuff from our study were identical to those reported pre- phocicerebrale, which has been implicated in seal mortality and viously15,20. The most abundant PS-OTU in both the chuff seal finger, a zoonosis among seal handlers34. Sequences closely (21.3%) and the blowhole (28.0%) was a member of the related to certain respiratory pathogens in dolphins, such as Cardiobacteriales, phylum Proteobacteria (Supplementary Data 3). M. elephantis, and , Based on data from both sequencing approaches, oral and were found in dolphin chuff and blowholes in low numbers. In gastric reads from sea lions were primarily assigned to the phyla addition, sequences belonging to the genus were Proteobacteria and Bacteroidetes (Supplementary Fig. 2). The found in many stomach and rectum specimens from both most abundant taxon in both oral and gastric microbiotas of sea dolphins and sea lions, most notably in the forestomach of lions was a Pasteurellaceae related to Actinobacillus porcitonsi. dolphins (53 of 56 specimens; 2.7% of all dolphin gastric PS Neisseria canis and several OTUs in Flavobacteria and reads); none were found in oral or respiratory specimens. Several Fusobacteria were also found in most of these specimens. The different clades within this genus were found, including rectal communities of sea lions were dominated by Bacteroidetes, H. cetorum, which has been previously reported in cetaceans Firmicutes and Fusobacteria. and pinnipeds30,35, with different Helicobacter FL sequence types found within the same dolphin forestomach specimen (Supplementary Fig. 11). In addition to sequences closely Microbiota from free-ranging dolphins. The composition of the related to H. cetorum, sea lion rectal specimens contained oral microbiota from wild, free-ranging dolphins in Sarasota Bay, sequences from a clade most closely related to Helicobacter canis; FL, was significantly different from that of MMP dolphins living in sequences of this clade have been isolated from a variety of San Diego Bay, CA (P 0.001 bootstrap test, Fig. 5 and o pinnipeds36. Because all animals included in our study were Supplementary Fig. 8). However, oral specimens were more similar healthy, the clinical relevance of these Helicobacter spp. for between wild and MMP dolphins than either were to specimens cetacean and pinniped health remains unclear. from the same oral sites of MMP sea lions. In contrast, the dif- ferences between the rectal communities of wild and MMP dol- phins did not reach statistical significance. Interestingly, Temporal stability of microbiotas at three body sites. To gain Fusobacterium OTU_C104, an abundant OTU in the oral micro- insight into the temporal stability of the marine mammal biota from dolphins, was nearly exclusive to MMP dolphins, microbiotas, seven MMP dolphins were sampled monthly over a whereas the abundance of Fusobacterium OTU_C718washigherin period of 5–6 months, with an additional sampling after 3 years. the mouths of free-ranging animals (Supplementary Fig. 10). In As with the human microbiota37,38, dolphin oral, gastric and addition, the oral microbiotas from MMP dolphins contained more rectal microbiotas were relatively stable, with intra-individual Cardiobacterium, Fibrobacteres and Synergistetes sequences than variation significantly lower than interindividual variation at each those of wild dolphins, while the rectal microbiotas from wild body site during the monthly sampling (P 0.001, Mantel test). dolphins contained more Campylobacteraceae sequences than those o The oral microbiota was the most stable and distinct between of MMP dolphins. However, because these two populations differ in animals on these timescales (Fig. 6). This stability appears to geographic location, food sources and medical care, the clinical decay on longer timescales; the greater similarity of specimens relevance of these differences remains unclear. sampled from the same dolphin was absent for oral and rectal communities and lower in gastric communities when specimens Pathogens in dolphin and sea lion microbiotas. None of the from the early monthly sampling period were compared with the bacterial pathogens typically associated with mortality in marine specimen collected 3 years later.

ab 1.5 Oral Rectal 1.0 Dolphin oral 1 Dolphin rectal 0.5 San Diego San Diego Alternate location Alternate location Sarasota Bay Sarasota Bay 0 0.0 Sea lion oral Sea lion rectal NMDS2 San Diego NMDS2 San Diego −0.5 Alternate location Alternate location −1 −1.0

−1 012 −2 −1 012 NMDS1 NMDS1

Figure 5 | Oral and rectal bacterial communities associated with bottlenose dolphins and sea lions. Oral (a) and rectal (b) bacterial communities associated with bottlenose dolphins and sea lions. NMDS Bray–Curtis ordination of PS reads in which colour shows specimen type and data point shape depicts location (San Diego or alternate location for the MMP animals, Sarasota Bay for the wild animals). Only one time point per animal is shown. A bootstrap test in which the wild/MMP status of the dolphins was resampled showed that the average Bray–Curtis distance within oral specimens from MMP and within wild dolphins was significantly lower than the average distance between oral specimens from MMP and wild dolphins (Po0.001). However, average Bray–Curtis distances of dolphin rectal communities were not statistically different within or between location groups (P ¼ 0.11).

6 NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516 ARTICLE

a bc Oral Gastric Rectal Animal ID 1.0 H 0.5 J 0.5 M 1 0.0 Q S 0.0 U NMDS2 NMDS2 −0.5 NMDS2 0 Y −0.5 3 year sample −1.0 FALSE −1.0 TRUE −0.5 0.0 0.5 1.0 1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 012 NMDS1 NMDS1 NMDS1

Figure 6 | Relative stability of dolphin-associated microbial communities. (a) Oral communities, (b) gastric communities and (c) rectal communities. NMDS Bray–Curtis ordination of PS reads for each body site individually for the seven dolphins sampled monthly for 5 or 6 months. Six of these were sampled 3 years later as well (open circles). For each of the three body sites, intra-individual specimens were significantly more similar than interindividual specimens during the monthly sampling period (Po0.001, Mantel test). This heightened intra-individual similarity decayed or disappeared entirely at three years, although conclusions are limited by the small number of animals.

Integrated view of marine and terrestrial mammal microbiotas. species in many marine census studies3,45. While terrestrial To examine whether host adaptation to life in the sea influences mammalian hosts are the subject of an ever-increasing number of the composition of host-associated microbiota, we compared oral, microbiome studies5,6,46 there have been relatively few studies of gastric and rectal communities of dolphins and sea lions to those marine mammal microbiotas. Here we present a large-scale of terrestrial and other marine mammals. First, we examined the molecular analysis of the taxonomic composition of the bacterial relationships among oral microbiotas from dolphins (this study), communities associated with different body sites in marine sea lions (this study), humans28 and dogs39. Bray–Curtis distance mammals from two different clades. We surveyed the oral, analysis showed that oral communities clustered according to rectal and gastric bacterial communities from dolphins and sea host species, with very little overlap in bacterial diversity between lions, as well as respiratory specimens from dolphins, using two host species (Supplementary Fig. 12). different sequencing strategies. Although the archaeal, eukaryotic To study the roles of host phylogeny, diet and habitat on the and viral components of the marine mammal microbiota composition of host-associated microbiota, we compared were not determined here, we found a remarkable amount the rectal microbiotas of sea lions, dolphins and manatees of previously unseen microbial diversity in these marine (this study), seals16, dugong24, river porpoises25 and polar bears40 mammals, substantial differences between the microbial with stool communities from 57 predominantly terrestrial communities of dolphins and sea lions, and distinctness of the mammalian species5,29,39,41–44 (Supplementary Data 4). microbiotas of marine mammals from those of the surrounding In general, the distal gut microbiotas from marine mammals, in aquatic habitat. particular those of dolphins and manatees, were distinct The microbiotas of dolphins in particular are characterized by from those of terrestrial hosts (adonis P ¼ 0.005) (Fig. 7a). high species richness, diversity and novelty. Representatives from Of note, the rectal microbiota of a dugong24, an herbivorous 30 bacterial phyla were found in the dolphin specimens. By marine mammal from the order Sirenia, clustered with those comparison, 22 phyla were found in a comparably sized data set of terrestrial Artiodactyl herbivores (Fig. 7b). Dolphin rectal of about 1 million reads from 6 body sites in humans37.A communities were most similar to those of bears, especially remarkably large number of bacterial phyla were found in the polar bears, which spend time in or near the sea, and to those dolphin oral (25 phyla) and gastric (22 phyla) microbiotas alone. of the hooded seal, river porpoise, and interestingly, the For comparison, sequences found in the human mouth28,47 and omnivorous hedgehog (Supplementary Table 2). The clustering stomach48 were affiliated with only 15 and 9 phyla, respectively. of the cetacean and bear gut microbiotas appeared to be driven by Several candidate phyla were found in dolphin oral and a lack of Bacteroidetes in both host groups (not shown). The respiratory microbiotas, such as GN02, H-178 and OD1. These rectal microbiotas of the carnivorous cetaceans were very distinct phyla are most often found in environmental samples49, and their from those of herbivorous Artiodactyls (Fig. 7c), although both consistent presence in mammal-associated bacterial communities groups fall within the same order Cetartiodactyla8,11. The sea lion is noteworthy. communities were most similar to those from the distal gut of Marine mammal body sites defined specific bacterial commu- three seal species and terrestrial members of the order Carnivora, nities with unexpectedly little overlap in OTUs. Our results are in including wolves and dogs (Supplementary Table 2). accordance with large-scale studies of the human microbiome Canonical correspondence analysis (CCA) revealed that that highlight the primary importance of body site in explaining dolphin microbiotas were distinct from other mammals (Fig. 8). variation in bacterial diversity37,50,51. In our study, host species Despite the still limited number of marine mammal microbiotas was also a strong determinant of community compositional available for analysis, these data are consistent with the variation, as illustrated by the distinctness of the dolphin and sea conclusion that living in the sea is a strong determinant of distal lion microbiotas, despite the MMP animals’ common habitat gut microbiota taxonomic composition. (seawater), location (San Diego Bay), and diet (identical fish and squid species, dietary supplements and food sources). We found that the oral, gastric and rectal communities of individual Discussion dolphins were relatively stable over time, similar to what has been Marine mammals, positioned at the top of the marine food web, found for dolphin respiratory samples20. We did not find any fulfil important ecological roles in the ocean and serve as sentinel patterns associated with age, sex or location.

NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516

b a

WD04 1 Mt07 Diet Carnivore 59 Omnivore Herbivore NMDS2 Mt08 24 0 1 92 40 Mt01 46 Mt09 Mt13 Mt03 89 55 Habitat −1 97 96 DolHH WD02 −1 0 1 2 91 WD01 Aquatic 93 57 62 NMDS1 5 47 70 61 WD09 Terrestrial NMDS2 9594 79 58 60 80 c 56 8 100 34 7 188177 4 0 2 12 10 36 Host order 50 644913 21 15526511 769 63 68 35 Carnivora 66 1914 6 3 1720 82 53 DolJ Cetartiodactyla 16 6778 90 101 1 87 73 Chiroptera 88 75 22 51 3799 43 1 84 83 25 Diprotodontia 85 86 26 7472 SL6 SL4 Hyracoidea 54 SL2 33 44 Insectivora 71 27 30 98 SL7 31 39 Lagomorpha 29 32 Monotremata 38

SL8 NMDS2 69 0 Perissodactyla SL5 23 41 Primates 28 42 Proboscidae −1 Rodentia 45 Sirenia Xenarthra −1 0 1 2 −1 NMDS1 −1012 NMDS1

Figure 7 | Comparison of distal gut communities obtained from different mammalian host species. (a) NMDS Bray–Curtis ordination of FL sequences from distal gut bacterial communities in 68 terrestrial and marine mammalian species, coloured according to host habitat. For these data sets, subsets of six individuals per host species (dolphins, sea lions and manatees) were chosen to provide a more balanced representation of marine mammal orders and diets. Specimens ‘100’ (dugong24) and specimens starting with ‘Mt’ (manatees, this study) were derived from herbivorous marine mammals (Order: Sirenia). Panels b,c show the same data as in a, coloured according to host diet (b) or mammalian phylogeny (Order) (c). See Supplementary Data 4 for more details on the individual data points and a key to the numbering shown in a. Supplementary Table 2 lists the most closely related communities to those from average dolphin and sea lion distal gut microbiotas.

Blautia Arthrobacter Hostgroup Clostridium Carnobacterium carnivora Collinsella Citrobacter Escherichia Distal gut 1 Halomonas Pseudomonas Akkermansia Sporosarcina Habitat Sarcina specimens Bacillus Streptococcus Bacteroides terrestrial Diet from mammal Bifidobacteria Comamonas Diet omnivore species Fibrobacter carnivore Helicobacter Hostgroup Lactobacillus primates Dolphin Prevotella Cetobacterium 0 Roseburia Dugong Ruminococcus Sphingobacterium Diet Manatee Peptostreptococcaceae herbivore Terrestrial Hostgroup cetartiodactyla Hostgroup Polar Bear −1 other Porpoise CC2: 4.1% Morganella Mycoplasmataceae Habitat Seal Mycobacterium Mycoplasmataceae aquatic Sea Lion Photobacterium Pseudoalteromonas Vibrio Taxa

−2 Campylobacteraceae Bradyrhizobium Desulfobulbus Pasteurellaceae Hyphomicrobium Paracoccus Variables Porphyromonas Sedimentibacter

−3 −5.0 −2.5 0.0 CCA1: 4.5%

Figure 8 | CCA of mammalian distal gut communities. Canonical Correspondence Analysis (CCA) plot showing the ordination of gut microbiotas from mammalian species (same as shown in Fig. 7, with subsets of six animals each for dolphins, sea lions and manatees) and bacterial taxa. Text in black displays the centroids of the three determinants that were tested, that is, habitat (aquatic and terrestrial), diet (carnivore, omnivore and herbivore) and host group (Carnivora, Cetartiodactyla, Primates and Other).

The dolphin forestomach is dominated by a previously clade, most dolphin forestomachs contain one or multiple unclassified bacterial species, FL-OTU A476, assigned to the Helicobacter sequence types. All dolphins in our study were phylum Tenericutes. Since it is abundant to the forestomach healthy without clinical signs of gastrointestinal disease; thus, the of most MMP dolphins in our study, and found in the distal gut clinical significance of this finding remains unclear. of wild dolphins and river porpoises25, we hypothesize that it The microbial communities found in the dolphin distal might be specific for delphinoids, and possibly involved in the gut differed from those found in other mammals in two digestion of fish. Metagenomic analyses might clarify the interesting ways. First, unlike in humans where faecal commu- metabolic properties of this bacterium. In addition to the A476 nities are more diverse than oral communities37, the rectal

8 NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516 ARTICLE communities of dolphins contained four times fewer bacterial of other dolphins. Sampling of mother and calf beginning at the species than their oral communities and added little PD to the time of birth, including vaginal samples of the mother, will be overall dolphin microbiota. Second, dolphin rectal communities valuable for elucidating the role of vertical microbiota inheritance contained strikingly and yet unexplained low numbers of in marine mammals, and early life assembly of their microbiotas. Bacteroidetes (o1%). Bacteroidetes are a major constituent of Since dolphins are extremely social animals, it is also possible that the distal gut microbiotas of most other mammals, including the microbial seeding occurs through physical contact between dugong24, a herbivorous sirenian and carnivorous pinnipeds, for animals (for example, play, mating, and dominance displays example, sea lions (this study and refs 16,22,23). The abundance such as raking)2. at this body site of Actinobacillus and Cetobacterium spp., both of While previous studies found that diet and host phylogeny are which have been described previously25,31,32, suggests that these both important determinants of the composition of the mammal- species are common members of cetacean gut microbiotas. associated microbiota5,6,46, our results suggest that habitat Although all respiratory specimens from dolphins contained (aquatic versus terrestrial) is also an important determinant. rich microbial communities, chuff contained higher phylum and The relatively distant ancestral relationships among extant species-level richness and diversity than the blowhole, suggesting marine mammals point to the importance of other shared that the deeper regions of the respiratory tract harbour more adaptations to life in the sea that may have shaped the structure, bacterial taxa than the upper regions, and probably provide a and presumably the function of the marine mammal microbiotas. better representation of the lower respiratory tract microbiome Further studies of bacterial, archaeal, eukaryotic and viral than the blowhole. This finding has implications for clinical diversity in a broader and larger selection of marine and management since chuff is also a less invasive sample type. terrestrial mammal species with differing functional capacity, Sea lion microbial communities were less diverse than those of lifestyle and environment will be useful for elucidating the dolphins. Their oral and gastric bacterial communities were underlying relationships between indigenous microbial similar in both phylum as well as species composition, and communities and their marine mammal hosts. clustered more closely than dolphin oral and gastric specimens. Understanding the composition of marine mammal micro- This finding is somewhat similar to the situation in humans, biotas in healthy animals and how they are affected by where gastric and oral communities share many bacterial environmental contaminants and changing prey profiles may species27. Similar to the distal gut communities of most provide valuable insight into potential long-term impacts of these mammals except dolphins, sea lion rectal communities increasing threats on marine mammal health. We expect that contained high numbers of Bacteroidetes and Firmicutes, and studies of this sort will help to define, predict and restore health low numbers of Proteobacteria. However, their rectal specimens in marine mammals13, and may provide a means for anticipating also harboured relatively high numbers of Fusobacteria, similar to the impact of environmental change on ocean ecosystems by what has been observed in seals16,22,23. allowing for the surveillance of microbiome health in these Rectal specimens from both dolphins and sea lions contained sentinel species. SFB (genus Candidatus Arthromitus). SFB are commensal bacteria anchored to the ileal mucosa and involved in the 52 Methods maturation of gut immune functions . Specific Arthromitus Animals included in this study Forty eight common bottlenose dolphins 33 . clades have been detected in different hosts , but they had not (T. truncatus) (age range 2–51 years; 23 males) and 18 California sea lions been detected in marine mammals prior to this study. Our study (Z. californianus) (age range 1–27 years; all male) were studied (Supplementary supports the host specificity of these indigenous intestinal Table 1). Thirty eight of the dolphins and all sea lions were managed by the US Navy MMP in San Diego, CA, USA. MMP dolphins and sea lions were housed in bacteria. open-water, netted enclosures in San Diego Bay. At the time of sampling, one Our study found a sharp boundary between microbial dolphin and four sea lions were housed at alternate locations in distant, distinct communities in marine mammals, their surrounding water and bodies of water. Both marine mammal species were fed mixtures of quality- their fish and squid diet. This finding is all the more dramatic controlled, frozen-thawed fish, including capelin (Mallotus villosus), herring (Clupea pallasii) and mackerel (Scomber japonicas), squid (Loligo opalescens) and since the sea water samples examined in this study were collected additional vitamin supplements (Vita-Zu Mammal Tablet # 5M26, Mazuri, immediately adjacent to each MMP and wild dolphin at the same Richmond, Indiana, USA). All MMP-managed animals that participated in this time as animal specimen collection, and since sampling sites such study were assessed to be healthy by a veterinarian (for example, no gastrointestinal as mouth and rectum of these animals are in constant direct diseases or overt clinical signs), and had not received antibiotics or other contact with seawater. Similar distinctions have previously been medications within 30 days prior to sampling. The MMP is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care found between earthworm-associated bacterial communities and (AAALAC) International and adheres to the national standards of the United 53 those of their soil environment , between the host-specific States Public Health Service Policy on the Humane Care and Use of Laboratory bacteria of marine sponges and those of their surrounding Animals and the Animal Welfare Act. As required by the US Department of seawater54, and more generally, between those of terrestrial Defense, the MMP’s animal care and use program is routinely reviewed by an Institutional Animal Care and Use Committee (IACUC) and the Navy Bureau of vertebrate gut microbiotas and those found in terrestrial Medicine and Surgery (BUMED). In addition, 10 free-ranging, long-term resident 55 environmental communities . It is also a sharper boundary bottlenose dolphins living in and near Sarasota Bay, Florida (South of Tampa Bay, than typically found in landscape ecology, where patch boundary in the Gulf of Mexico) were captured, examined, sampled and safely released by the transitions are often more gradual56. Strong selective pressure Chicago Zoological Society’s Sarasota Dolphin Research Program, conducted under exerted by the specific structures, conditions and immune the National Marine Fisheries Scientific Research Permit No. 552–1785 (ref. 3). surveillance mechanisms at mammalian mucosal surfaces may explain our findings. Yet, the marine mammal–seawater Specimen collection. Collection of biological specimens is common practice in boundary is not absolute. Very low numbers of dolphin- marine mammal veterinary care, and follows the guidelines set forth in the Chemical Rubber Company (CRC) Handbook of Marine Mammal Medicine2. Oral associated OTUs were found in seawater, leaving open the specimens were obtained from dolphins and sea lions by swabbing the gingival possibility that marine mammals might be colonized in part sulcus of the lower jaw with a sterile Fisherbrand Polyester-Tipped Applicator through contact with seawater. (Fisher Scientific, Pittsburgh, Pennsylvania, USA). Gastric fluid was collected by A more likely source of microbial seeding is contact between passing a small half inch outside diameter PVC stomach tube (JorVet, Jorgensen Laboratories, Loveland, Colorado, USA) into the forestomachs of MMP dolphins mother and calf. Our study included specimens from three and stomachs of MMP sea lions, applying mild suction, then slowly removing the mother–calf pairs but all three calves were Z2 years, and their tube and draining the contents into a sterile 120 ml specimen container (Parter communities did not cluster more closely to the adults than those Medical Products, Carson, California, USA). The sample was then transferred into

NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516 a sterile 15 or 50 ml polypropylene centrifuge tube (Corning, Corning analysis. Of these, 2,681 sequences were obtained from 6 seawater samples, 992 Incorporated, Corning, New York, USA) prior to freezing. Rectal mucosa swab from 7 fish and squid specimens, 13,184 from dolphins (22 oral, 5 gastric and 19 specimens were collected via sterile swab (same brand as above) passed through the rectal specimens), and 3,173 from sea lions (6 oral, 6 gastric and 6 rectal anal orifice. Seawater was collected alongside sampled animals (MMP and specimens). FL data sets consisted of an average of 260 sequences (range: 92–673) free-ranging) at the time of animal sampling in a sterile 120 ml specimen container. per specimen. Sequences were assigned to bacterial phyla using the SILVA Water samples were also transferred into sterile 50 ml polypropylene centrifuge release 102, Ref-NR ARB database62 and the Ribosomal Database Project (RDP) tubes prior to freezing. Two types of respiratory specimens were collected from Release 10 Classifier63. OTUs (FL OTUs) were defined using a 99% sequence MMP dolphins: a blowhole mucosa swab (same brand as above) and a ‘chuff’ similarity cutoff, using a 1,256- or 684-nucleotide mask on the PCR products specimen, for which the animal was signalled to forcefully exhale onto an 87 mm generated by reverse primers 1391R or 806R, respectively, filtering out the filter (Protran BA85, GE Healthcare Whatman, Dassel, Germany) placed above the hypervariable regions. The 99% cutoff in this setting roughly corresponds to blowhole. Specimens of fish (capelin, herring and mackerel) and squid, all of species-level groupings. The FL sequence OTU table is available as Supplementary human-consumption quality, used to feed MMP dolphins and sea lions, were also Data 5. Phylogenetic neighbour-joining trees were created using the Jukes Cantor collected. Fish and squid remained frozen at 20 °C until they were shipped on correction and the 1256-nt filter. Additional alpha- and beta-diversity analyses dry ice for processing. All other specimens were immediately placed under were performed with the R package phyloseq64. The R code is available in refrigeration, then processed and stored at 80 °C, and later shipped on dry ice for Supplementary Software 1. subsequent processing. Determination of sequence and taxa novelty. The percentage of novel OTUs DNA extraction. Seawater samples were prepared for DNA extraction by filtering found in the FL sequence data set, using different levels of sequence identity, was the sample through a 150 ml 0.2 mm Nalgene analytical filter unit (cellulose nitrate determined by comparing the sequences of OTU representatives to published filter, 47 mm diameter; Thermo Scientific, Waltham, Massachusetts, USA). After sequences using BLAST. Novelty of the OTUs from dolphin and sea lion oral, filtration, the filter was cut into pieces in a sterile Petri dish using a sterile scalpel, gastric and rectal FL data sets (this study) was compared with the novelty of and placed into a sterile 2-ml screwcap vial (Sarstedt, Nu¨mbrecht, Germany). human oral sequences28, human gastric sequences27, and human rectal biopsy Gastric fluid (1.5 to 2 ml) was centrifuged for 2 min at 15,000g, and the pellet was and stool sequences29 at their time of publication. These previously published data used for DNA extraction. Fish and squid specimens (one whole animal for herring, sets were chosen since they were all amplified and analysed by nearly identical mackerel and squid, and about 100–200 gram of capelin pieces) were homogenized methods. All FL sequence sets were rarefied to 1,013 sequences and 6 individuals in 50 ml PBS in a Stomacher paddle blender (Seward Limited, West Sussex, UK) per host species, except the human intestinal samples, for which 6 specimens using two rounds of 60 s each on medium speed. Oral and rectal swabs were from a total of 3 individuals were used, and the human gastric samples for processed directly for DNA extraction using the QIAamp DNA mini kit (Qiagen, which 15 individuals were chosen. For each FL OTU obtained in this study, a Valencia, CA; tissue protocol), according to instructions by the manufacturer, and BLAST search was performed using the standard parameters and excluding the then eluted in 200 ml AE buffer (supplied in the kit). During every extraction round, primer positions, and the percentage identity of that OTU to its top hit in the NCBI empty tubes (about 1 per 10 specimens) were processed in parallel to serve as nt database (version March 2015) was used as the outcome. For the human negative extraction controls. sequences, the BLAST identity percentages obtained at the time of publication were used. Fish identification. The species identity of the fish used to feed the MMP dolphins and sea lions was confirmed by amplification and sequencing of the mitochondrial Comparing oral microbiotas from different mammalian species. FL oral 16S cytochrome B gene using primers CytB-F, 50-AAAAACCACCGTTGTTATTC 0 0 0 rDNA sequences derived from 12 MMP and 10 wild dolphins (7,149 reads) and 6 AACTA-3 (ref. 57), and CytB-R, 5 -CGICCTCAGAAKGAYATTTGICCTCA-3 MMP sea lions (1,050 reads) obtained in this study were compared with FL 16S (modified from ref. 57). One microlitre of the extracted fish DNA was added to an rDNA sequences from dogs and humans. The canine oral microbiota data set39 amplification reaction with 20 ml 2.5 HotMasterMix (5 Prime Inc., Gaithersburg, consisted of 4,253 sequences from 51 dogs (294 OTUs); only sequences from MD, USA) and 0.4 mM of each primer, with a final volume of 50 ml. The PCR broad-range bacterial amplification reactions were included in our analysis. The programme consisted of an initial denaturation at 94 °C for 2 min, 35 cycles human oral microbiota data set28 included 10,285 sequences obtained individually of 95 °C for 30 s, 55 °C for 30 s, 65 °C for 30 s and a final incubation for 65 °C for from 10 humans with healthy mouths (241 OTUs). A neighbour-joining tree 8 min. PCR products were purified using the QIAquick PCR purification kit containing representatives of the 1,465 OTUs found in the combined data set was (Qiagen), and Sanger-sequenced (Sequetech, Mountain View, CA). Comparison of built using Halobacterium salinarum as the outgroup. The 89 microbial the sequences using BLAST confirmed the fish species used for feeding the marine communities were compared with each other using the Bray–Curtis distance and mammals to be M. villosus (capelin), C. pallasii (Pacific herring) and S. japonicus NMDS ordination in the phyloseq R package64. The R code is available as (chub mackerel), with 100% sequence identity to Genbank sequence entries Supplementary Software 2. DQ457494, KC588467 and KM2000011, respectively.

Comparing gut microbiotas of terrestrial and marine mammals Clone library construction and Sanger sequencing. To obtain near-FL (about . FL rectal 1,400 bp) 16S rRNA gene sequence information, the extracted DNAs from dolphin, sequences from 9 MMP and 10 wild dolphins and 6 MMP sea lions (this study) sea lion, fish, squid and seawater specimens were amplified using broad-range were compared with published distal gut data sets derived from 66 terrestrial and marine mammal species from other studies. First, sequences derived from stool bacterial-specific 16S rRNA gene primers, 8FMB (a 10:1 mixture of primers 5 8FM, 50-AGAGTTTGATCMTGGCTCAG-30 and Bifidobacterium spp.-specific specimens of 57 different terrestrial mammals were downloaded from SILVA 4 primer 8FB, 50-AGGGTTCGATTCTGGCTCAG-30) and 1391R (50-GACGGG release 102 (http://www.arb-silva.de); only sequences with a Pintail value 75% 0 58 29 were selected to obtain a data set of 19,335 sequences. In addition, we included CGGTGTGTRCA-3 ) . PCRs were performed as described previously , using the 25 following conditions: 5 min at 95 °C, 25 cycles of 30 s at 94 °C, 30 s at 55 °C and 90 s sequences obtained from faeces of a Yangtze finless porpoise , colon contents from grey seals, harbour seals and hooded seals16, pooled polar bear faeces40, the at 72 °C, followed by 8 min at 72 °C. No amplification product was observed in 44 65 24 extraction controls and negative PCR controls. Because of eukaryotic background faeces of a giant panda , colon of domestic dogs , faeces of a dugong , caecal contents of lean mice41 and faeces of wolves43. We also included stool specimen amplification issues, fish, squid and some wild dolphin specimens were amplified 29 0 sequences from three healthy humans and stool specimen sequences from three using primer mixture 8FMB (see above) and 806R (5 -GGACTACCAGGGTAT 42 CTAAT-30)27, using a 30 s extension time and yielding an 800-bp fragment. After healthy humans prior to antibiotic exposure . Finally, to add more representatives pooling four replicate amplification reactions, PCR products were purified from from the aquatic herbivorous Order Sirenia, we added sequences from rectal agarose gels using the QIAamp Gel Purification kit (Qiagen). Clone libraries were samples of wild manatees (Florida) obtained by PS. The total data set, including constructed with the TOPO TA cloning kit (Invitrogen, Carlsbad, CA). Escherichia sequences obtained in this study, consisted of 46,321 16S rRNA sequences derived coli TOP10 (Invitrogen) transformants were lysed and cloned inserts were from 68 different mammalian species (135 individuals) (Supplementary Data 4), amplified using M13 primers as directed in the cloning kit. PCR products were and was analysed using the Quantitative Insights Into Microbial Ecology (QIIME) software package66. In short, OTUs in the combined mammalian gut data set were purified using the MultiScreen PCRm96 Filter Plate (EMD Millipore, Billerica, MA), and sequenced from both ends using the T3 and T7 primers and Sanger sequencing picked against a reference database using a similarity threshold of 97% and the UCLUST package67, and OTU representatives were aligned and assigned to technology (J. Craig Venter Institute, Rockville, MD). Forward and reverse reads 61 were assembled using the Sequencher program (Gene Codes, Ann Arbor, MI). taxonomic groups against the Greengenes core set alignment . OTUs were grouped into genera, and an NMDS ordination on Bray–Curtis distance was constructed with phyloseq64 and used to explore beta-diversity. CCA was run on Phylogenetic analysis based on FL sequences. Assembled FL 16S rRNA OTU tables using the vegan R package and plots were constructed displaying gene sequences were aligned with the online Greengenes NAST aligner59 specimens, OTUs, and centroids representing the relative contribution of the (http://greengenes.lbl.gov), and 20,852 sequences were uploaded into the environmental variables (habitat, diet and host ) considered as Greengenes-version of the ARB database60,61. The alignment was further perfected factors. To provide a more balanced representation of the marine mammal by manual optimization. A total of 462 chimeras (2.2%) and 360 poor-quality reads orders, subsets of six dolphins and six manatees were included in the CCA; (1.7%) were manually identified and removed from the analysis, leaving 20,030 all six sea lions were kept in this analysis. The R code is available as Supplementary high-quality, non-chimeric sequences derived from 77 specimens in the final Software 3.

10 NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516 ARTICLE

PCR amplification for PS analysis. Extracted DNA from specimens derived from different-dolphin averages among monthly samples. Because of the small number 36 dolphins and 18 sea lions was used for PS. A 589-bp fragment of the of animals and the single late sample, we make no significance statements about bacterial 16S rRNA gene spanning the hypervariable V3–V4–V5 regions was this comparison. These calculations were performed in R, and are available in amplified. The forward primer (50-CGTATCGCCTCCCTCGCGCCATCAG-NN Supplementary Software 4. NNNNNNNNNN-GC-ACTCCTACGGGAGGCAGCA-30) contained the 454 Life Sciences primer-A sequence, a unique 12-nt error-correcting Golay code (a different barcode for each specimen; depicted by the Ns in the sequence)68,a Shared OTUs in different specimen types. Shared OTUs between dolphin, sea two-base linker sequence (‘GC’) and the broad-range bacterial primer 338F. lion and seawater were identified using a rarefied PS data set; data were selected The reverse primer (50-CTATGCGCCTTGCCAGCCCGCTCAG-AA-CCGTC from 18 (12 MMP and 6 wild) dolphins (120,527 reads), 18 sea lions (116,167 AATTCCTTTGAGTTT-30) contained the 454 Life Sciences primer-B sequence, a reads) and 18 seawater samples (13 samples taken alongside MMP dolphins or sea two-base linker (‘AA’) and the broad-range bacterial primer 906R. One to 2.5 mlof lions and 5 alongside wild dolphins; overall 69,377 reads). All three data sets were extracted DNA were added to a 25 ml amplification reaction containing rarefied to 69,377 reads in QIIME66. The number of OTUs that were shared HotMasterMix and 0.4 mM of each primer. The PCR programme consisted of an between or unique for specimen types was plotted in a Venn diagram. Fish initial denaturation at 94 °C for 2 min, 25 cycles of 95 °C for 30 s, 53 °C for 30 s, specimens were not included in the analysis because not all fish used to feed the 65 °C for 30 s and a final incubation at 65 °C for 8 min. Three or four replicate PCR MMP animals yielded PS reads, and because diet samples were not available for mixes were pooled, and the amplicons were analysed on agarose gels. In case of wild dolphins. weak amplification reactions, an additional 4 replicate PCRs (30 cycles in cases of very low yield) were performed and pooled. PCR amplicons were purified using the QIAquick PCR purification kit or the UltraClean-htp PCR clean-up kit Quantitative PCR. To analyse the presence of specific bacterial groups that were (MoBio, Carlsbad, CA). DNA concentrations were determined using the abundant in FL sequence libraries derived from particular specimen types but were Quant-iT High Sensitivity DNA Assay kit (Invitrogen). Equimolar amounts of the absent or rare in FL data sets from other habitats, four specific qPCRs were barcoded PCR products were pooled, ethanol precipitated and gel-purified developed. In short, qPCRs were developed for the detection of OTU_O614, using the QIAquick Gel Extraction kit (Qiagen). The purified pool was sequenced P. ubique, abundant in seawater; Fusobacterium-like OTU_C104, abundant in using the 454 Life Sciences Genome Sequencer FLX Titanium platform dolphin oral specimens; Tenericutes sp. OTU_A476, abundant in dolphin (Roche, Branford, CT). forestomachs; and OTU_B031, abundant in dolphin rectal specimens. Primers and probe sequences are provided in Supplementary Table 3. All qPCR assays were performed in 20 ml reactions containing 1 TaqMan Universal PCR Master Mix Processing and analysis of PS reads. A total of 337 specimens and 26 extraction (Applied Biosystems), 0.9 mM of each primer, 0.2 mM of the probe and 2 mlof controls from this study were analysed in three PS runs. Reads were processed extracted DNA. The thermal cycling programme consisted of 95 °C for 10 min, 66 using the QIIME software package . In short, reads from each sequencing run followed by 40 cycles of 95 °C for 30 s, 55 °C for 30 s, 60 °C for 45 s, 65 °C for 15 s, were quality-filtered (quality score 25, length 4200 and o1,000 nt, no ambiguous 72 °C for 30 s. All reactions were carried out in triplicate on a StepOnePlus Real bases, homopolymer length o6), assigned to their original source specimen using Time PCR System (Applied Biosystems). Ten-fold serial dilutions of known the unique barcodes and primer sequences were removed. Denoising was quantities of plasmids containing the specific sequence target for each assay were performed in specimen type-specific batches of roughly 10,000 reads each, using used to generate standard curves. Sequence abundance was calculated using SDS the denoise_wrapper.py command within QIIME. Chimeras were identified and software version 2.1 (Applied Biosystems). removed using UCHIME67 using a customized reference database derived from the Greengenes 12–10 release clustered at 99% sequence identity threshold (available on request); 14.7% of reads were removed by this step. Sequences were clustered References into OTUs using a similarity threshold of 97% against the same reference database 1. Estes, J. A. et al. Trophic downgrading of planet Earth. Science 333, 301–306 as used for UCHIME and allowing for new clusters. OTU-representative sequences (2011). were aligned and masked using the Lane mask, and a phylogenetic tree was built 2. Dierauf, L. A. & Gulland, F. M. D. CRC Handbook of Marine Mammal using the FastTree software implemented in QIIME. Taxonomy was assigned to Medicine (CRC Press, 2001). each OTU-representative sequence using the RDP classifier, a curated Greengenes 3. Wells, R. S. et al. Bottlenose dolphins as marine ecosystem sentinels: developing reference database (see above), and a confidence score of at least 80%. In the data a health monitoring system. Ecohealth 1, 246–254 (2004). set used in this study, 8,117 OTUs were found among 915,150 reads (including 26 4. Evans, P. G. H. & Raga, J. A. Marine Mammals: Biology and Conservation extraction controls, which combined yielded 180 reads). (Springer, 2001). The PS data OTU table is available as Supplementary Data 6. Not counting the extraction controls, the average number of reads per specimen was 2,715 5. Ley, R. E. et al. Evolution of mammals and their gut microbes. Science 320, (range 0–14,184). 1647–1651 (2008). 6. Muegge, B. D. et al. Diet drives convergence in gut microbiome functions across mammalian phylogeny and within humans. Science 332, 970–974 Analysis of bacterial diversity from PS data. Bacterial diversity within speci- (2011). mens (alpha-diversity) or between specimens (beta-diversity; using Bray–Curtis 7. Foote, A. D. et al. Convergent evolution of the genomes of marine mammals. 66 64 distances) was calculated using QIIME and the R package phyloseq , and a non- Nat. Genet. 47, 272–275 (2015). rarefied data set from samples with 372 reads or more. The R code for the PS data 8. Meredith, R. W. et al. Impacts of the Cretaceous terrestrial revolution and KPg set is available as Supplementary Software 4. extinction on mammal diversification. Science 334, 521–524 (2011). 9. Geisler, J. H. & Theodor, J. M. Hippopotamus and whale phylogeny. Nature Calculation of gain in PD. The gain in PD is the branch length a sample adds to a 458, E1–E4 (2009). tree containing other samples. It is the branch length exclusive to that sample69,70. 10. Uhen, M. D. Evolution of marine mammals: back to the sea after 300 million The gain in PD depends on the other taxa already present in the tree, so the order years. Anat. Rec. (Hoboken) 290, 514–522 (2007). in which samples are added plays an important role. PD gain was calculated in 11. Spaulding, M., O’Leary, M. A. & Gatesy, J. Relationships of Cetacea QIIME66 by using the ‘unifrac_g_full_tree’ parameter executed in the (Artiodactyla) among mammals: increased taxon sampling alters beta_diversity.py script. Using the fraction of branch length in the full tree that is interpretations of key fossils and character evolution. PLoS ONE 4, e7062 exclusive to a particular specimen, it calculates the PD on cumulative OTU tables, (2009). where each specimen is added to the calculation in a user-defined order. For the 12. Flynn, J. J., Finarelli, J. A., Zehr, S., Hsu, J. & Nedbal, M. A. Molecular exploration of marine mammal-associated bacterial diversity, two different orders phylogeny of the carnivora (mammalia): assessing the impact of increased of sample addition were tested. sampling on resolving enigmatic relationships. Syst. Biol. 54, 317–337 (2005). 13. Venn-Watson, S., Smith, C. R. & Jensen, E. D. Primary bacterial pathogens in Temporal stability of bacterial communities in dolphins. To test the temporal bottlenose dolphins Tursiops truncatus: needles in haystacks of commensal and stability of bacterial communities within the same animal, seven dolphins were environmental microbes. Dis. Aquat. Organ. 79, 87–93 (2008). sampled monthly for 5 or 6 months at the gastric, oral and rectal anatomical sites. 14. Nelson, T. M., Apprill, A., Mann, J., Rogers, T. L. & Brown, M. V. The marine Six of these animals were sampled again 3 years later. Pairwise Bray–Curtis mammal microbiome: current knowledge and future directions. Microbiol. distances were calculated between all specimens from the same anatomical site, Aust. 36, 8–13 (2015). based on PS reads. A Mantel test was used to determine whether samples from the 15. Johnson, W. R. et al. Novel diversity of bacterial communities associated with same dolphin were more similar than were samples from different dolphins during bottlenose dolphin upper respiratory tracts. Environ. Microbiol. Rep. 1, 555–562 the monthly sampling period. To be specific, a Mantel test was performed on the (2009). matrix of Bray–Curtis distances and the distance matrix obtained by assigning 0 to 16. Glad, T. et al. Ecological characterisation of the colonic microbiota in arctic and pairs of samples from the same dolphin and 1 to pairs of samples from different sub-arctic seals. Microb. Ecol. 60, 320–330 (2010). dolphins. The late sample (after 3 years) was treated separately: The average 17. Buck, J. D., Wells, R. S., Rhinehart, H. L. & Hansen, L. J. Aerobic distance between late samples and monthly samples was calculated for samples microorganisms associated with free-ranging bottlenose dolphins in coastal gulf from the same and different dolphins, and compared with the same- and of Mexico and Atlantic ocean waters. J. Wildlife Dis. 42, 536–544 (2006).

NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516

18. Schaefer, A. M., Goldstein, J. D., Reif, J. S., Fair, P. A. & Bossart, G. D. 48. Delgado, S., Cabrera-Rubio, R., Mira, A., Suarez, A. & Mayo, B. Microbiological Antibiotic-resistant organisms cultured from atlantic bottlenose dolphins survey of the human gastric ecosystem using culturing and pyrosequencing (tursiops truncatus) inhabiting estuarine waters of charleston, SC and Indian methods. Microb. Ecol. 65, 763–772 (2013). River lagoon, FL. Ecohealth 6, 33–41 (2009). 49. Brown, C. T. et al. Unusual biology across a group comprising more than 15% 19. Morris, P. J. et al. Isolation of culturable microorganisms from free-ranging of domain Bacteria. Nature 523, 208–211 (2015). bottlenose dolphins (Tursiops truncatus) from the southeastern United States. 50. Huttenhower, C. et al. Structure, function and diversity of the healthy human Vet. Microbiol. 148, 440–447 (2011). microbiome. Nature 486, 207–214 (2012). 20. Lima, N., Rogers, T., Acevedo-Whitehouse, K. & Brown, M. V. Temporal 51. Segata, N. et al. Composition of the adult digestive tract bacterial microbiome stability and species specificity in bacteria associated with the bottlenose based on seven mouth surfaces, tonsils, throat and stool samples. Genome Biol. dolphins respiratory system. Environ. Microbiol. Rep. 4, 89–96 (2012). 13, R42 (2012). 21. Apprill, A. et al. Humpback whale populations share a core skin bacterial 52. Gaboriau-Routhiau, V. et al. The key role of segmented filamentous bacteria in community: towards a health index for marine mammals? PLoS One 9, e90785 the coordinated maturation of gut helper T cell responses. Immunity 31, 677– (2014). 689 (2009). 22. Smith, S. C., Chalker, A., Dewar, M. L. & Arnould, J. P. Y. Age-related 53. Thakuria, D., Schmidt, O., Finan, D., Egan, D. & Doohan, F. M. Gut differences revealed in Australian fur seal Arctocephalus pusillus doriferus gut wall bacteria of earthworms: a natural selection process. ISME J. 4, 357–366 microbiota. FEMS. Microbiol. Ecol. 86, 246–255 (2013). (2010). 23. Nelson, T. M., Rogers, T. L. & Brown, M. V. The gut bacterial 54. Webster, N. S. et al. Deep sequencing reveals exceptional diversity and community of mammals from marine and terrestrial habitats. PLoS One 8, modes of transmission for bacterial sponge symbionts. Environ. Microbiol. 12, e83655 (2013). 2070–2082 (2009). 24. Tsukinowa, E. et al. Faecal microbiota of a dugong (Dugong dugong) in 55. Ley, R. E., Lozupone, C. A., Hamady, M., Knight, R. & Gordon, J. I. Worlds captivity at Toba Aquarium. J. Gen. Appl. Microbiol. 54, 25–38 (2008). within worlds: evolution of the vertebrate gut microbiota. Nat. Rev. Microbiol. 25. McLaughlin, R. W., Chen, M. M., Zheng, J. S., Zhao, Q. Z. & Wang, D. Analysis 6, 776–788 (2008). of the bacterial diversity in the faecal material of the endangered Yangtze finless 56. Cadenasso, M. L. et al. An interdisciplinary and synthetic approach to porpoise, Neophocaena phocaenoides asiaeorientalis. Mol. Biol. Rep. 39, 5669– ecological boundaries. Bioscience 8, 717–722 (2003). 5676 (2012). 57. Russell, V. J. et al. Use of restriction fragment length polymorphism to 26. Sanders, J. G. et al. Baleen whales host a unique gut microbiome with distinguish between salmon species. J. Agric. Food Chem. 48, 2184–2188 (2000). similarities to both carnivores and herbivores. Nat. Commun. 6, 8285 (2015). 58. Palmer, C., Bik, E. M., Digiulio, D. B., Relman, D. A. & Brown, P. O. 27. Bik, E. M. et al. Molecular analysis of the bacterial microbiota in the human Development of the human infant intestinal microbiota. PLoS Biol. 5, e177 stomach. Proc. Natl Acad. Sci. USA 103, 732–737 (2006). (2007). 28. Bik, E. M. et al. Bacterial diversity in the oral cavity of 10 healthy individuals. 59. DeSantis, Jr T. Z. et al. NAST: a multiple sequence alignment server for ISME J. 4, 962–974 (2010). comparative analysis of 16S rRNA genes. Nucleic Acids Res. 34, W394–W399 29. Eckburg, P. B. et al. Diversity of the human intestinal microbial flora. Science (2006). 308, 1635–1638 (2005). 60. Ludwig, W. et al. ARB: a software environment for sequence data. Nucleic Acids 30. Goldman, C. G. et al. Novel gastric helicobacters and oral campylobacters are Res. 32, 1363–1371 (2004). present in captive and wild cetaceans. Vet. Microbiol. 152, 138–145 (2011). 61. DeSantis, T. Z. et al. Greengenes, a chimera-checked 16S rRNA gene database 31. Foster, G. et al. Actinobacillus delphinicola sp. nov., a new member of the and workbench compatible with ARB. Appl. Environ. Microbiol. 72, 5069–5072 family Pasteurellaceae Pohl (1979) 1981 isolated from sea mammals. Int. J. Syst. (2006). Bacteriol. 46, 648–652 (1996). 62. Quast, C. et al. The SILVA ribosomal RNA gene database project: 32. Foster, G. et al. Cetobacterium ceti gen. nov., sp. nov., a new gram-negative improved data processing and web-based tools. Nucleic Acids Res. 41, obligate anaerobe from sea mammals. Lett. Appl. Microbiol. 21, 202–206 D590–D596 (2013). (1995). 63. Wang, Q., Garrity, G. M., Tiedje, J. M. & Cole, J. R. Naive Bayesian classifier for 33. Yin, Y. S. et al. Comparative analysis of the distribution of segmented rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl. filamentous bacteria in humans, mice and chickens. ISME J. 7, 615–621 (2013). Environ. Microbiol. 73, 5261–5267 (2007). 34. White, C. P. & Jewer, D. D. Seal finger: A case report and review of the 64. McMurdie, P. J. & Holmes, S. phyloseq: an R package for reproducible literature. Can J Plast Surg 17, 133–135 (2009). interactive analysis and graphics of microbiome census data. PLoS ONE 8, 35. Harper, C. G. et al. Helicobacter cetorum sp. nov., a urease-positive e61217 (2013). Helicobacter species isolated from dolphins and whales. J. Clin. Microbiol. 40, 65. Suchodolski, J. S., Camacho, J. & Steiner, J. M. Analysis of bacterial diversity in 4536–4543 (2002). the canine duodenum, jejunum, ileum, and colon by comparative 16S rRNA 36. Harper, C. G. et al. Isolation and characterization of novel Helicobacter spp. gene analysis. FEMS Microbiol. Ecol. 66, 567–578 (2008). from the gastric mucosa of harp seals Phoca groenlandica. Dis. Aquat. Organ. 66. Caporaso, J. G. et al. QIIME allows analysis of high-throughput community 57, 1–9 (2003). sequencing data. Nat. Methods 7, 335–336 (2010). 37. Costello, E. K. et al. Bacterial Community Variation in Human Body Habitats 67. Edgar, R. C., Haas, B. J., Clemente, J. C., Quince, C. & Knight, R. UCHIME Across Space and Time. Science 326, 1694–1697 (2009). improves sensitivity and speed of chimera detection. Bioinformatics 27, 38. Lozupone, C. A., Stombaugh, J. I., Gordon, J. I., Jansson, J. K. & Knight, R. 2194–2200 (2011). Diversity, stability and resilience of the human gut microbiota. Nature 489, 68. Hamady, M., Walker, J. J., Harris, J. K., Gold, N. J. & Knight, R. 220–230 (2012). Error-correcting barcoded primers for pyrosequencing hundreds of samples in 39. Dewhirst, F. E. et al. The canine oral microbiome. PLoS ONE 7, e36067 (2012). multiplex. Nat. Methods 5, 235–237 (2008). 40. Glad, T. et al. Bacterial diversity in faeces from polar bear (Ursus maritimus) in 69. Faith, D. P. Conservation evaluation and phylogenetic diversity. Biol. Conserv. Arctic Svalbard. BMC. Microbiol. 10, 10 (2010). 61, 1–10 (1992). 41. Turnbaugh, P. J., Baeckhed, F., Fulton, L. & Gordon, J. I. Diet-induced obesity 70. Lozupone, C. A. & Knight, R. Global patterns in bacterial diversity. Proc. Natl is linked to marked but reversible alterations in the mouse distal gut Acad. Sci. USA 104, 11436–11440 (2007). microbiome. Cell. Host. Microbe. 3, 213–223 (2008). 42. Dethlefsen, L., Huse, S., Sogin, M. L. & Relman, D. A. The Pervasive Effects of an Antibiotic on the Human Gut Microbiota, as Revealed by Deep 16S rRNA Sequencing. PLoS Biol. 6, e280 (2008). Acknowledgements 43. Zhang, H. H. & Chen, L. Phylogenetic analysis of 16S rRNA gene sequences We thank all trainers and veterinarians at the US Navy Marine Mammal Program in San reveals distal gut bacterial diversity in wild wolves (Canis lupus). Mol. Biol. Rep. Diego for their help in obtaining the specimens used in this study, and David Barton, 37, 4013–4022 (2010). Alvin Chang, Eric Chow, Christina Rohlik and Koji Yasuda for laboratory technical 44. Zhu, L. F., Wu, Q., Dai, J. Y., Zhang, S. N. & Wei, F. W. Evidence of cellulose assistance. This research was supported by Office of Naval Research grants N00014-07-1- metabolism by the giant panda gut microbiome. Proc. Natl Acad. Sci. USA 108, 0287, N00014-10-1-0233 and N00014-11-1-0918 (D.A.R.), NIH R01 GM086884 (S.P.H. 17714–17719 (2011). and B.J.C.), NSF DMS 1162538 (S.P.H. and B.J.C.), and the Thomas C. and Joan M. 45. Veldhoen, N., Ikonomou, M. G. & Helbing, C. C. Molecular profiling of marine Merigan Endowment at Stanford University (D.A.R.). Samples from wild dolphins in fauna: integration of omics with environmental assessment of the world’s Sarasota Bay, FL, were obtained primarily through the support of Dolphin Quest, Inc., oceans. Ecotoxicol. Environ. Saf. 76, 23–38 (2012). and Morris Animal Foundation’s Betty White Wildlife Rapid Response Fund. These 46. Delsuc, F. et al. Convergence of gut microbiomes in myrmecophagous samples could not have been collected without the assistance of Dr Jay Sweeney, mammals. Mol. Ecol. 23, 1301–1317 (2014). Dr Deborah Fauquier and the staff, students and volunteers of the Sarasota Dolphin 47. Wade, W. G. The oral microbiome in health and disease. Pharmacol. Res. 69, Research Program. We thank Robert K. Bonde from the Southeast Ecological Science 137–143 (2013). Center, US Geological Survey, for obtaining manatee samples.

12 NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10516 ARTICLE

Author contributions Competing financial interests: The authors declare no competing financial interests. E.M.B., R.S.W., K.P.C., E.D.J., S.V.-W. and D.A.R. designed the research. E.M.B., E.K.C., B.J.C., S.P.H. and D.A.R. performed the experiments and analysed the data. E.M.B., Reprints and permission information is available online at http://npg.nature.com/ E.K.C., A.D.S., B.J.C., R.S.W., K.P.C. and D.A.R. wrote the manuscript. reprintsandpermissions/

How to cite this article: Bik, E. M. et al. Marine mammals harbour unique microbiotas Additional information shaped by and yet distinct from the sea. Nat. Commun. 7:10516 doi: 10.1038/ Accession codes: All high-quality, FL 16S rRNA gene sequences obtained in this study ncomms10516 (2016). have been deposited in the GenBank Nucleotide database with accession codes JQ191150 to JQ217071 and KC257481 to KC260956. The 16S rRNA gene pyrosequencing data have This work is licensed under a Creative Commons Attribution 4.0 been deposited in the NCBI’s BioProject database with accession codes PRJNA174530 International License. The images or other third party material in this (oral, gastric, rectal and water sequences) and PRJNA279427 (respiratory and additional article are included in the article’s Creative Commons license, unless indicated otherwise water sequences). in the credit line; if the material is not included under the Creative Commons license, Supplementary Information accompanies this paper at http://www.nature.com/ users will need to obtain permission from the license holder to reproduce the material. naturecommunications To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

NATURE COMMUNICATIONS | 7:10516 | DOI: 10.1038/ncomms10516 | www.nature.com/naturecommunications 13